American trademark law has long operated on the assumption that there exists an inexhaustible supply of unclaimed trademarks that are at least as competitively effective as those already claimed. With respect to word marks in particular, the conventional wisdom holds that we will always enjoy a surplus of preexisting words, and in any case trademark adopters can simply coin new words, the supply of which is thought to be effectively “infinite.”
This empirical assumption — that the supply of good, competitively effective trademarks is inexhaustible — has long formed the foundation of important theoretical conjectures at the core of trademark law and policy. The most significant of these is that when we grant exclusive rights in a trademark, the cost to competitors, consumers, and more generally to the public domain is inconsequential.
Contrary to the conventional wisdom in trademark law, however, popular media has lately begun to make the opposite empirical claim: that the supply of good trademarks is, in fact, exhaustible and that we have very nearly exhausted it. For example, the New York Times recently asserted that “[a]lmost every naturally occurring word has been claimed, which is why namers so often arrive at portmanteaus (Accenture derives from ‘accent’ and ‘future’) or drop vowels (Flickr and Tumblr) or change letters (Lyft).”
The article recalled an entrepreneur’s description of his efforts to find a name for his new company: “Every name we liked, either somebody already had it or it wasn’t trademarkable or it meant something pornographic in another language.” For the Chicago Tribune, the focus was craft beer and the headline was “Craft Beer Makers Running Out of Names. How About Flip Donkey Doodleplunk?” NPR has further reported that “[v]irtually every large city, notable landscape feature, creature and weather pattern of North America — as well as myriad other words, concepts and images — has been snapped up and trademarked as the name of either a brewery or a beer.” For The Guardian, the focus was band names under the headline “FKA Twigs, Slaves, Deers: Are We Running Out of Band Names?” The article observed that “[a]ll the best monikers have been taken, and now the lawsuits are flying.” Popular television series have also taken up the theme. Futurama and South Park have each featured scenes in which nearly all words or word combinations have already been trademarked. Meanwhile, free speech advocates have grown increasingly vocal about the pervasive trademarking of everyday words. The YouTube duo the Fine Brothers announced in 2016 that they had applied to register the word “react,” after their series of videos. The public reaction was critical and merciless. One commenter joked about registering the word “the” and threatened that “anyone who says it get[s] sued.” The commenter was no doubt unaware that at the time there were already eleven active trademark registrations claiming just the word THE. Another commenter stated simply: “REACT is not yours to trademark.” In response to the furor, the Fine Brothers withdrew their trademark application. To the extent that legal and popular commentary has engaged the question of the exhaustibility of the supply of trademarks, the discussion has been based at best on anecdata and at worst on raw assertion. This Article seeks to move beyond both by systematically studying all 6.7 million trademark applications filed at the U.S. Patent and Trademark Office (PTO) from 1985 through 2016 together with the 300,000 trademarks already registered at the PTO as of 1985, which is made possible by the PTO’s recently released Trademark Case Files Dataset. By contrast, trademark congestion is the process by which an already-claimed mark is claimed by an increasing number of different trademark owners. This Article focuses specifically on word marks and thus on word-mark depletion and word-mark congestion. Overall, the data show that the conventional legal wisdom is wrong and the conventional popular wisdom is right. The supply of word marks that are at least reasonably competitively effective as trademarks is finite and exhaustible. This supply is already severely depleted, particularly in certain sectors of the economy, and levels of depletion continue to rise. Those marks that are registered are growing increasingly congested. The result, as the data reveal, is that new trademark applicants are increasingly being forced to resort to second-best, less competitive marks, and the trademark system is growing increasingly — perhaps inordinately — crowded, noisy, and complex.
Specifically, the data present compelling evidence of substantial word-mark depletion, particularly with respect to the sets of potential marks that businesses prefer most: standard English words, short neologisms that are pronounceable by English speakers, and common American surnames. Together with the PTO dataset, we use the Corpus of Contemporary American English dataset of the 100,000 most frequently used words in American English and the U.S. Census’s list of the 151,672 most frequently occurring surnames in the United States to show the extraordinarily high proportion of English words and common surnames that are already registered as trademarks. We further show the remarkably low proportion of words and surnames not confusingly similar to already-registered marks. With respect to short neologisms, we use the Carnegie Mellon University Pronouncing Dictionary and LOGIOS Lexicon tool to construct a phonetic representation of each word mark applied for or registered in the PTO dataset. Based on these data, we show that the supply of short neologisms not confusingly similar to already-registered marks is substantially declining. Finally, because many trademark applicants prefer to be able to register any new mark as a domain name in the .com top-level domain, we use Verisign’s .COM Zone File consisting of some 128 million currently registered domain names in the .com top-level domain to illustrate the near-total depletion in that space of standard English words, common American surnames, and short neologisms.
Given these conditions, new applicants are increasingly resorting to suboptimal marks. The data indicate that applicants are applying less often for standard English words and common surnames and more often for more complex marks, as measured by character, syllable, and word count. We think that applicants are modifying their conduct in this manner primarily to avoid applying for marks that the PTO would refuse to register on the basis of section 2(d) of the Lanham Act, which denies the registration of a mark that, due to its similarity with an already-registered mark, would confuse consumers as to source. Yet applicants appear to be increasingly unsuccessful in avoiding such refusals. We use our original dataset of all 2.1 million trademark office actions issued by the PTO from 2003 through 2016 to report the increasing rate at which the PTO is refusing applied-for marks on the basis of section 2(d). Despite these trends, one class of applicants appears to be doing fine. Incumbent applicants (those applying based on previous registrations) continue to apply for non-neologisms at a rate substantially higher than nonincumbent applicants and continue to enjoy very low section 2(d) refusal rates.
The data also reveal compelling evidence of substantial word-mark congestion. Consistent with increasing section 2(d) refusal rates, trademark applicants are increasingly resorting to what we term “parallel registrations.” Two firms can use exactly the same mark provided that their uses would not confuse consumers as to source (for example, DELTA for faucets and DELTA for airlines). Nevertheless, a trademark owner would prefer to be at best the only firm in the economy using a particular mark and at least the only firm in its economic sector doing so. Parallel uses may not confuse consumers as to source, but each use destroys the uniqueness and blurs the distinctiveness of the other, particularly for newer entrants. They also increase consumer search costs. Yet the data show steady increases in parallel registrations of frequently used English words and common surnames both across and within classes of goods and services. Firms appear to be increasingly settling for sharing marks with others.
These findings urge a rethinking of many of the fundamental assumptions underlying trademark law. Most importantly, they emphasize that the granting of trademark rights imposes real costs on the ecology of the trademark system, and that as we begin to test the limits of this ecology, these costs are mounting. New market entrants face significant barriers to entry in the form of the cost of searching for an unclaimed mark and in the ongoing cost of using a less effective mark. Consumers must cope with an ever more crowded field of trademarks consisting of increasingly complex marks that may refer to multiple different sources. The public domain must cope with the fact that, as we report below, when we use our language, about three-fourths of the time we are using a word that someone has claimed as a trademark.
These findings also counsel fundamental reforms of trademark law and doctrine. For example, given the costs a trademark registration imposes on the rest of the trademark system, particularly when the registration consists of a desirable word like a standard English word, we could enforce the use requirement more aggressively, as the PTO has already begun to do. We could also elevate the required showing of secondary meaning that an applicant must make when seeking to register a descriptive mark. In general, we could institute various schemes of congestion or peak pricing with respect to application, maintenance, and renewal fees to compel registrants to internalize more of the costs that their registrations impose on competitors, consumers, and the public domain. We could also take the degree of trademark depletion and congestion in particular sectors into account in the protectability, infringement, and trademark fair use analyses. The challenge in all cases will be to ensure that these reforms do not impose even greater costs on entrants. To be sure, many of these reforms may strike current trademark owners as unthinkable. But they are unthinkable only if we continue to fall back on the conventional wisdom that the trademark system is based on an inexhaustible resource, or in any case, that our economy could never reach a stage of development that would begin to test the limits of this resource. As we show, this conventional wisdom is wrong, and we must begin to consider ways of adapting to the limits of the trademark system.
Part I provides background on the trademark registration process and addresses the question of how to define the universe of good trademarks. Of course, the supply of possible trademarks, like the supply of possible personal names, is theoretically infinite. At the extreme, new firms could simply adopt alphanumeric codes of indefinite length to identify themselves. We explain why this is the wrong way to think about the universe of potential trademarks and why marketers rightly see this universe in very different terms. Part II describes our datasets. Parts III and IV present evidence of word-mark depletion and word-mark congestion, respectively. Though we draw upon “big data,” our evidence takes the form of straightforward descriptive statistics showing clear trends over time. Part V sets out the implications of our findings for trademark law and policy.